Empirical data from two studies address the clinical validity of discrepancy criteria for identification of children with developmental language disorders (DLD). Study 1 involved 256 preschoolers clinically defined as DLD and meeting inclusionary criteria for normal hearing, intellectual, neurological, and psychiatric status. Application of alternative psychometrically derived discrepancy criteria identified only 40% to 60% of the clinically defined group as language disordered. Study 2 applied nonverbal IQ-language performance discrepancy criteria to 368 eight-year-old, randomly selected control subjects, resulting in over 45% of the controls being identified as DLD. Factors contributing to underidentification in Study 1 and overidentification in Study 2 are discussed, raising questions regarding the validity of discrepancy criteria for identification of DLD children.